LogIn
Self-taught 3d artist Adam Kenyon showed how he created this stunning tribute to ‘Last of Us’ in just over 5 weeks.
Adam Kenyon has recently presented his fan art inspired by a Naughty Dog’s masterpiece “The Last of Us”. It is a huge UE4 scene full of tiny details that perfectly capture the spirit of the post-apocalyptic novel. He spent more than 5 weeks to complete the project and the results are amazing. We present to you our exclusive detailed breakdown of this scene.
Introduction
Hello everyone at 80 Level!
Firstly, I’d like to thank with my deepest appreciation to 80.lv for inviting me to do this interview, also thank you for the great feedback from you guys on my project and the community of Artstation, it means a lot to me. This is actually my first time being interviewed about my work. My name is Adam Kenyon and I am from a small town in the South East of the United Kingdom called Gosport, near Portsmouth. For the last five years, I have mostly self-taught and developed myself as a 3D Environment and Props artist, posting projects up on Artstation and receiving important critique advice from the community. It’s a really great and diverse community of talent.
Around the age of 7 or so, I took a keen interest in gaming and technology. My first PC was an IBM 386 and I started playing games such as Commander Keen, Hexen, Wolfenstein and DOOM. Then later on I got into consoles as well such as Megadrive/SNES/N64/Saturn/Playstation but I don’t think I got interested in making games until around the age of 14 when I had my first Pentium 3 system and Windows XP. I started playing around with software such as The Games Factory and Multimedia Fusion (if anyone remembers those! haha). It was my first stepping stone I believe in game development and I had big dreams and ambitions of making the next best game, I even dreamed I was working at SEGA or Nintendo who were my big idols of the time, I never knew that I would become a games artist today as originally I wanted to be a programmer. Going all the way back to high school days, I eventually developed a huge interest in traditional art such as drawing, painting and also IT Technologies. I took GCSE Art and GCSE Technology as some of my final studies as I thought these subjects would be of some use to getting into the games industry.
Then going on from school I chose to do a BTEC in Art and Design at Fareham College and I also did a BTEC in ICT (Computing course for practitioners). On the Art course I also studied GCSE Photography; I wanted to learn more about real world objects and to study and draw them. All of this education eventually enabled me to go the University of Portsmouth where I actually ended up doing BCs Hons in Computing, but I actually didn’t enjoy it as much as I thought, even though I completed the course, I felt that I was not a programmer and that my passion actually lied in Art. I spent a few years afterwards working part time jobs until I had the idea of wanting to go back to Portsmouth University, and I managed to raise the funds to do so. So I studied BSc Computer Games Technologies for two years. Unfortunately, I did not gain a full degree due to financial circumstances to do the final year but I did gain a Higher National Diploma in the course.
Whilst I was at this course, I did some paid quality assurance testing work for a company called Arooga Games; this was my first time working in a studio. I was testing their software for bugs on a mobile platform and writing reports. One of my projects in University was called Business Enterprise where I helped develop and publish a 2D iPhone game called GOGORocket! I was responsible for the 2D artwork such as backgrounds and sprites. After studying at the University of Portsmouth, I was trying to look for opportunities to gain a foothold in the games industry.
With no experience and still much to learn about the art of game development, I was struggling to find work. It was until one day when I was on Facebook, a friend of mine told me about a friend of his was looking for artists to help develop a game that was already in development for a little while. I got in contact with Mark Silvester, Josh Baker and James Wilkins who asked me to do a small art test. I successfully passed the art test by making some small assets. The indie company is called Quarter Circle Games and is primarily based in Bournemouth but most of the work is done over Skype as we are scattered all over the place. The game is a first person survival horror game called The Peterson Case and I contribute mostly environment props for the game which runs on the Unity 5 game engine. The Peterson Case was posted on Steam Greenlight and was successfully greenlit in under a week. Currently the game has been in development for over two years and has gone through many transformations as technology and skills got better.
Recently we acquired a publisher and investor which is one of the toughest obstacles to overcome when in independent game development. We are hoping to publish the game later in 2017 but this is to be confirmed yet. You can see a trailer for the game here.
I got my first real studio gig last year at Turbulent Designs which is a simulations company who were looking for two artists that had experience with Substance tools and Unity 5. The company was in a transition of getting into the games industry side. Working on a game called ‘Homicide Detective’ for an American client who was fund raising for various charitable causes. We were tasked with designing and producing the whole game within six months, around three months into development, the game subsequently was cancelled for reasons I cannot disclose. I mostly worked on designing and developing assets for the environment and occasionally dressing them. My employment was terminated due to redundancy after six months of working there. I really enjoyed my time there and worked with some fantastic people.
The Last of Us Fan Art
There are many reasons why I wanted to do this project. Firstly, The Last of Us, what an amazing game! I had a PS3 for quite some time but weirdly enough I had never played it until I owned a PS4 and then I had to get the re-mastered version to see it for myself. When I first played it, I was amazed at the excellent storytelling, the post-apocalyptic worlds enriched with detail and the characters that are so intriguing. I had always been quite the fan of Naughty Dog’s work prior to The Last of Us such as Uncharted and I just knew this game was one of those unique experiences and it really was! Over the last year or so I had constantly pondered about doing a TLOU styled environment. It was until recently that I saw the amazing work of Romain Dauger of Ubisoft fame, that it triggered an inspiration in me. I went to the internet to search for Naughty Dog conceptual art work and soon I had learnt of the work from John Sweeney who was responsible for TLOU concept art. Romain’s Arstation post on his tips and tricks was a huge help to me in this project, so I thank Romain once again for his help.
There were many scenes I could have chosen but I chose to do the house concept he had drawn. It’s somewhere in the game but I couldn’t remember where as I haven’t played the game in a while now. The artwork is nothing short of phenomenal and this got me thinking, it’s a huge environment and that is going to give me a good challenge and I do love a good challenge. The project would give me a scope that was bigger than I could have imagined and that’s what I was aiming for. This project would push me to model things that I have never touched on before. Things such as animal sculpting, intricate wood details and multi texture vertex painting, foliage etc. So I guess you could say my main aims were to push my skills higher, to show a diverse range of materials and models and to go beyond anything I have achieved before. It’s a real satisfying feeling when you complete a rather big project and yet it turns out better than you thought. For references, there was only one main reference and that was the original concept art view from John Sweeney except for upstairs which was originally going to be made as well! I wanted to create something realistic and atmospheric. I think that I managed to achieve my vision.
However, doing upstairs as well, would of taken me a great deal of time to do on top of what I have already made downstairs. It may be something I will do later on when I have time to do so and expand that project. So, I decided to scale the scope of the project down on just concentrating on the downstairs only, putting all my energy into that. I basically, remembered everything that was great about the Last of Us and looked at images from the game, the kind of props and styles of texturing. Took those ideas and poured it into the melting pot of possible props by making a list out of things to be modelled, gathering as much references as possible. The house concept showed me that it had so much character, so many different elements to it that it excited me to be making it. There are a few easter eggs in the environment as I will show in the screenshots. One of them is Joel and Ellie in the detail golden picture frame. An old guitar I had made a while ago, I actually kind of based it on the guitar you see in TLOU Part 2 trailer that Ellie is playing, so that’s another one. On the walls is some firefly emblem decals as the fireflies might have been there. There quite a few more I would have added probably.
Production
The production was a mix of both those methods you described. At first it all started in 3DS Max with a dirty block out. To save time and not have to go remodel everything, I tend to model what will be the final primitive object such as walls, floor tiles, coving etc. and I will unwrap them as well. These pieces are put in place in MAX to give some kind of rough idea how it will go together, everything is fully modular. This is a bit of a slow process but it does give me the right dimensions and ensures I get everything to the scale I need. The stairs is a bit tricky (get more into that later in the modeling question) because it intersects with the ceiling and has an opening. Not all the walls could use the same UV’s like I wanted because of door gaps and corners so that was a bit of a limitation. It looks like I tried to build the whole environment in 3DS Max but it’s not the case. I think everyone has a different way of doing things but in the case of this environment, building a rough block out will definitely help you to understand the size and dimensions and corner parts that need to be aligned correctly.
Panels on the walls and architrave pieces would be reused over and over by slicing up the model and keeping the same UV’s. Same with the door arches, bannister wood slats, windows, ceiling parts. Anything that was modular would be unwrapped earlier on so that I could make the tiling process easier.
I had also modeled a few assets and textured them to get an idea of the style and look of the environment such as the dresser cabinet at the far back by the stairs, doing a few tests in an early UE4 scene as shown in the image. I was daunted at where to start on this project, you can really start anywhere and just go for it but I chose to start building out the stair case as I believe that sets the tone for this particular scene.
Starting with the stairs helped me to find out how high the ceiling was and then the walls just fell right in place, so if the walls and arches were not the right size I could just easily adjust them in 3DS Max and reimport in UE4.
Why Unreal Engine 4?
As to why using the UE4 games engine, I am quite experienced with it and it makes integrating PBR workflow so amicable. Various reasons include the blueprint node system. It’s a great engine; it’s a really powerful engine. It allows artists to easily import assets by dropping and dragging them into the world. Artists can quickly and effortlessly setup materials and drop them on to unwrapped assets. In essence, UE4 is great for rapid prototyping. It helps artists to visualize their content and provides unlimited design possibilities. The rendering is really nice and so is the lighting system. Unreal allows me to create a quick and simple blueprint to create wind, to blow my foliage about. It also has vertex painting, which allows me to paint things like dirt and roughness to a floor, a wall or blend textures such as height and normal maps to create damage. I feel UE4 is more powerful in terms of rendering than Unity 5, which is still a great engine as well but it just generally has more features such as post processing effects built in and better looking lighting. Bear with some of the issues UE4 has at times (such as stability) and you can build really nice things in UE4, adding foliage and multi material vertex painting to blend textures together and many other things such as particles.
Modeling
As there is quite a bit of modeling in my environment, I shall try and break it down to the meat and potatoes of the modeling process. All the base meshes and unwrapping took place in 3DS Max. High poly models were detailed and sculpted in Zbrush. I used Max also to do any retopology.
Walls
The walls are pretty simple geometries; they are just planes with extrusions. Most of it is semi modular with some parts sharing the same UV maps to keep draw counts down and performance up. Despite looking fairly simple, this proved to be a bit of a challenge to get most of the wallpaper texturing of the same size and required a lot of experimentation. The walls are not all the same though, with some of the walls requiring gaps for windows to fit in and doorways. For the texturing, most of it was done in Photoshop. I used a wallpaper texture sample that was as close as possible to the original, and tiled it in Photoshop over the top of my UV’s.
I add painted on dirt and stain marks. This was a repeated process for all the wallpaper meshes. One room had slightly different coloured wallpaper, same pattern, so changing hue and brightness values changed this.
Floor
The floor tiles are also quite a simple process. It’s basically one single plane tile, unwrapped in max. Then creating a box to the length and width of the plane, and making it look like a wooden plank. I then shift+drag out copies of the plank to around 10 copies, making sure the planks fill the plane, leaving a gap either side for tiling. You can either build these planks outside of the plane then centre align or just align them yourself directly on the plane. Either way, you have to make sure the planks are directly on top of the plane to take into zbrush and detail the planks. The idea behind this is to bake these planks straight on to the plane so we can get Normal, Cavity, AO and Height information.
I then exported both the low poly plane with the planks on top to Zbrush for detailing but I shall go into more detail in the later question about materials process.
Wall panels
The wall panels are modeled in a very similar fashion to the floor tiles, they too are also modular. The panels have some rounded bar parts on them and I wanted to make sure all of that was on one map and that all the panels shared the same map. The corner parts were a complete pain; I’d say that was quite a challenge for me. The answer to this issue and tiling was to model the long panel by the stairs and also model on to it a corner part with the chamfered corner. After unwrapping, I could slice up the geometry how I wanted for long and short parts such as in-between window frames and wall ends. All I did for the under the stair case lining was to do a diagonal quick slice, and deleted out the part I didn’t need and it retained the UV’s. So I modeled the panel like so:
Staircase and balcony
The stair case was box modelled. I first started making one step, unwrapped it so I could make modular steps ascend in UE4. I blocked the steps layout out in MAX so I could model the handrail around this. I did the same method for the bannister poles and supports. The side panels on the side of the stair case would be used later for the wallpaper texture going up the side. This just acted as a template whilst I was modeling the staircase.
I wanted to add some damage/detail to the stair case steps and add a damaged banister pole. So I took these parts into Zbrush for some detailing.
Getting the staircase to handle rail to align in UE4 was tricky, I had some initial issues with either too many stair case steps or too little, otherwise it would be taller than the ceiling. Bear in mind, this is the first and only stair case I have built for a game environment going up to an upstairs, it seems simple and it is but it was a bit awkward getting it right at first.
Also making sure that the wallpaper meshes don’t collide with the stairs, trying to make it all look seamless took some time. I think the most (and will re-illiterate this later on in the interview) time spent on the environment was tiling materials and models. There is a lot of modular tiling going on.
Ceiling and flakes
The ceiling is on two maps and modelled from standard planes. The first one is for the main hallway and is on one 2048×2048 map, the other is one plane split into two parts sharing the same UV’s on another 2048×2048 map. Initially, I had serious trouble establishing a flaked ceiling look. I made some flakes in Photoshop for foliage tool, and it worked well on the floor however I couldn’t get the foliage to paint on the ceiling very well and ended up scrapping this idea. Later on in development, it was the awesome ysPaintPeel filter by Yuri Snitko that solved the problem.
Windows and curtains
The windows are all modular and box modeled so I won’t go over too much detail about them but the curtains are made from using a highly tessellated plane and cloth simulation. The plane is around 100 by 100 segments. It is then converted to an edible poly. At the top edges I use an edge ring to select every other two edges, I then extrude these out into a curled shape (this is where the collision happens). A rail that holds the curtain on is used as part of the collision. Using an auto key, I selected all the top vertices of the curtain and scaled them in whilst recording an animation. When you simulate, this will push the curtain fabric in together, forming frills in the top of the curtain and high poly creases in the fabric.
Enriching the Environment
I agree, there are many different assets that add character and soul to the environment, which was the initial idea to make everything as interesting as possible. I will try and break down as much as possible because there may be too much to cover here, so I shall try covering the most important aspects. There is much more I wanted to add but didn’t want to be too overkill. I added in a lot of things on the ground, such as broken splintered wood, bits of shrapnel, debris, plaster flakes from the ceiling, leaves and metal pipes. Most of the splintered wood, shrapnel and debris was generated using an awesome little script, that is free, it’s called Debris Maker 2 by Aaron Dabelow. It’s a fantastic tool, and it really helped speed up the development process by generating modular high poly models such as broken wood and metal. It has so many options! I re-optimized these high poly parts for any of the debris generated suitable for UE4 using Pro Optimizer carefully and manually editing some parts before baking down.
Deferred decals are also quite important if you want to add that little bit extra. As you can see in the environment I have mould, moss, entrenched holes in the walls. The broken plaster decals are an illusion using a normal map and AO. Things like graffiti and dripping dirt are a nice touch to the abandoned styled of this project.
Bricks were made by me in Zbrush and then retopoed in 3DS Max like so. I used a few different brushes, such as clay build-up, dam standard, cracks orbs brush, mallet, trim dynamic, move, and standard. The bricks were an idea by myself to show decay and brittleness of walls, which went nicely with the smaller modular generated pieces of debris. The way I have laid everything out is to make the environment look as messy and chaotic as possible.
Foliage I shall talk about in the next question of how I worked with the foliage and made the different plants. The foliage does make up for a lot of floor space in certain areas of the environment to make it seem nature has reclaimed the house back. Looking at the original reference, you can tell that the owner of the house was perhaps into hunting game, and as you can clearly see that was a huge challenge because it required me to model a deer head. It was a lot more difficult than I had thought as I haven’t done much animal sculpting but I had to tackle it one way or the other. I had to have it in my environment.
I also added a hunting shotgun on an old driftwood rack, which was an original idea by myself and only further adding to the theme and personality of the people who lived in that house. The deer head had to be sculpted first in Zbrush, from Zspheres and then slowly getting the forms down which was also very tricky as it took me quite a while just to get the basic shapes right. Then moving on to the models eyes, mouth, nose and eye socket areas. The antlers were also made with Zspheres.
And the final render in Zbrush of the high poly model upclose. Afterwards, the meshes are then retopoed in 3DS Max.
Rendered in Marmoset Toolbag.
Next, I will talk about the shotgun on the rotten driftwood mount. This was an important aspect as you enter the environment and in-between the two deer heads above one of the arches. I thought back to films like Evil Dead, I had this idea that if you’re going to have deer heads in this environment, you might want to show the weapon used to hunt them, portray it as trophy to reflect the character of dwellers that once lived in the environment. I think it provides a story.
The gun and the mount were both modeled in 3DS Max. The mount was done by using a reference of an old double barrel shotgun, mounted onto some driftwood. The driftwood I drew around using splines, then converted to edible poly. I then joined up the vertices and using the border, I extruded out the edge to thicken it out. The shoe horns are also done by spline modeling.
The guns meshes consist of mostly cylindrical, box and manual plane modeling. The oddly shaped gun butt was a little tricky because the shape curves in and out with some sharp edges. I think it took me about two tries to get the butt looking right because the guns body melts into that pointed shape in the middle of the butt. I mixed a cylinder with some manual modelling and then welding up bits here and there and moving vertices around to get the shape right.
After I unwrapped everything including the gun, I sent the models to Zbrush for detailing.
I shall briefly discuss the furniture of this environment that I have made particularly the Piano, Piano Stool, Chesterfield Chair and the Jacobean table. These are all quite detailed pieces of furniture that add style and substance to the environment. There is also a guitar, which is reference to Ellie from Last of Us 2 trailer.
Originally I wanted a grand piano, where the final piano is right now, however, I felt that it was going to be way too big and wouldn’t meld with the environment so I went with this Piano.
I felt that this was a really nice looking Piano and it had some really nice alphas I could make in Photoshop for Zbrush. The white keys was a simple box that I made and I made some high poly keys in 3DS Max to bake down on to the box so it would keep the Piano lower in polygons. Whilst, the black keys were very low poly box modelled keys. Most of the piano is box modelled and had details done in zbrush like scratches, scuffs and patterns which were made in Photoshop for alphas. I used projection master to drag these out.
The Piano stools legs are modelled with a rough block out, sculpted in ZBrush and then sent back to 3DS Max for retopology. Sometimes I find this workflow easier to do that trying to model some organic in Max first. I used some of the built in alphas and dam standard to detail the legs and middle section. The beads are quite simple to do as well, using radial symmetry (mesh had its pivot centred before export) and dragging out on the Y axis, creating multiple beads using an alpha. I used the standard brush to create some folds and droops in the leather.
The Chesterfield chair and Old English Jacobean table were modeled in 3DS Max from reference, then taken into Zbrush for detailing. All the furniture is modeled in a similar workflow. Not too much retopology is required for these pieces of furniture as were baking to hard surfaces. The Chesterfield sofa required some manual shapes to be created using a plane and extruding/welding. The chair was retopologized to get the forms in the low poly and get nice clean bakes. The Jacobean table is a completely modular model. The drawers, legs, supports, and panels are all the same UV’s. The top is not split though; just either side of the table is a modular kit as are the drawers.
For the foliage there was going to be two types, foliage that would be procedurally painted using the foliage tool in UE4 and some of the bigger plants to be manually placed. The foliage does more than just painting plants though, as my first time using it, it can paint pretty much any static mesh you throw at it. So the foliage tool was also used to paint flakes on the floor, bits of debris, nuts and bolts and leaves. Speedtree wasn’t actually used that much, I used the free trees pack to place some trees outside the windows. I can model trees but this vastly saved a lot of time. I only have the free licence for Speedtree not the premium one, so I had to make use of what I had to hand.
The foliage is super low poly as you can see from the renders in Marmoset. They all consist of a normal map and an albedo. I give each plant some slight roughness in Unreal engine by hooking up a constant. I got some nice free plant textures from textures.com, created alphas from them using colour range in Photoshop and setup some simple plant materials. The way they are modeled is just with planes, manipulated here and there with a few vertices pushed in and out to make them look more natural. On the big plant for example I used bend modifier in different directions and rotated many copies. Moving the leaves about and bending them is an efficient way to make the plant come to life. Same technique with the other plants, just rotating, bending, moving vertices and moving parts about. I used the super powerful Knald to generate my normal maps; it’s really easy, quick and efficient.
The foliage tool is also very quick and easy to use in UE4. I just dropped and dragged all my static meshes into the foliage tool; setup a few parameters such as radius and density. At first, I started with the grass strands, wanted to test the waters and establish some vegetation in corners of the environment. Eventually I started adding some flowers in to break up the repetitiveness of the grass meshes. I made the grass thicker by increasing the density, although you have to be careful, too much and it will cause serious overlapping, so you have to set a good distance as well. The other meshes such as the bigger plant and ivy, were hand placed so I could get them into grooves, cracks and the front door way.
Materials
I don’t really have a set way per say of approaching the materials. I tend to try and analyze the reference first, looking at how the textures would or should look like. Taking in the information I read from the surface or characteristics of the reference for the asset I have yet to create. I will show an example from my work to try and better explain how I go about creating the texture workflow, the floor tiles for example.
In the below image, is a sample of the floor tile high poly asset in Zbrush. I need to get normal, height, AO and cavity maps from this. I use a few different brushes to get the effect I want. Using Trim dynamic to add damage to edges, orbs cracks brushes to add cracks, grains and splits. I also used a few free wood alphas/brushes from Fredo Guiterrez from Artstation. Thanks dude!
Then I baked out the High poly model to the low poly model using Substance Painter. When we bake out with Substance Painter, we get some very useful maps including the powerful curvature map. The Curvature Map allows us to trace any dents, grooves and edges. This makes it a lot easier to paint masks where we want things like dirt and moisture. I use curvature maps on nearly everything and also cavity maps, which really bring out those crevices in the wood. All these different kinds of maps you can use can really make a difference to your texturing way before you even start using Substance Painter materials.
Substance Painter is a fantastic revolutionary tool in any CG Artists arsenal. Both Painter and Quixel Suite have revolutionized the way we present texturing. I prefer Substance tools though; there is a lot more freedom. Painter feels right at home with Photoshop and gives you the flexibility to paint your model in so many ways. So many awesome features such as built in alphas, particle system, procedural masks and filters. It really speeds up your workflow and allows you to visualize your PBR creation before it gets in to the engine, a versatile and easy to you use tool for anyone. I still use Photoshop as well in conjunction with SP in order to do any detailed Albedo maps, ID maps, blending textures together or creating alphas/decals.
Here in the picture above you can see the texture that I painted in Spotlight from Zbrush. I take advantage of the curvature map and use it to trace over the grooves, cracks and splits so I can paint a mask over where I want the old dust to be. I also use a dirt brush in Painter with some roughness values to make it look wet and rotten in places. This is part of the wear and tear process but most of the wear and tearing details comes from connecting with the high poly first. I make sure that I am getting a good read on the damaged information whilst sculpting passes, ensuring I am satisfied with the end result before baking.
Using brushes such as Trim Dynamic allows me to beat up models or chip away at them. I sometimes use the spray function, if I am working on something metal, so I can create dents. I also use a variety of built in alphas to add smaller details. Cracks orbs brushes are by far the best way to do scratches and cracks; it gives you a lot of control from softness to hardness. Clip curve or clip rectangle is also a great way to chop up parts and chip bits off. Height maps that pickup on these details can be immensely helpful in pushing the visual information. The Height map can be used in UE4 as this mesh is tessellated, which means I can make the boards look raised and less flat whilst also making those dents and grooves pop out.
I also make use of different layers in substance painter, filters and height control. I like to manually go in and make use of the height control in Painter, such an awesome feature that I can paint normals on with such freedom.
Here is an example of the floor material in UE4; it shows how the flat tessellation is setup and the substance painter materials. By the way, I am no pro master shader expert either. I shall also show my moss and dust vertex painting material.
There were some certain issues that I came across with material creation for UE4 such as using too much tessellation. This would cause the floor boards to frail and it is something that I had forgotten to fix up in the final video render. You just have to experiment with the constant value plugged into your world displacement to get it just right. Other things such as the carpets in the main reference I was using. I was not too sure how to go about making realistic fur in the beginning. I came across a great tutorial by Barry Lowndes, who taught me how to use Parralax Occlusion and some Substance Designer, which is also new to me. It was incredible how it turned out; it looked like realistic carpet fur and it also solved the issue with fur on the toy bear. Parralax Occlusion is a type of mapping that adds depth and definition to a displaced surface. Because it was an instanced material I had created, I could easily change all the values without having to re-edit the main material nodes. Learning Designer is a steep learning curve and I only learnt a small part of it but I yearn to learn more. The wallpaper was a bit of a pain to get it be consistent throughout the environment, I had to do a lot of manual Photoshop work for that. Because I had lots of different walls I had to try and keep the wallpaper at the same scale for each wall mesh. I used a variety of stained Photoshop brushes to get the old moldy, moisture dirt look. Vertex painting was also new to me in this project, especially multi blended materials, so a lot self-teaching on my part.
Rendering
The scope and scale of this project was huge, really huge. Probably the biggest environment I have ever worked on to date. The lighting of this environment took a long time. I used mostly static lighting for better performance; they were easier to work with as well. I also used some spot lights as well, especially the light shafts coming through the window for soft touches and atmopshere. That was one of the very first things I wanted to tackle and it too was a new thing to me. The light shaft is basically a cylinder with an emissive map on it but also a panner node to animate the dusty light and a depth node to fake the gradient of the light. I was aiming for realism in this environment and that was the main emphasis. I was trying to create an atmosphere that was surreal, that was mysterious and yet with the nature overtaken the place, beautiful. Spotlights helped to emphasize light rays coming through the windows and the front door, which has a gap open on purpose.
The particles floating around were created to enhance the feel and life of the environment. Dusty particles and spores were an idea I had mid development cycle. I wanted the light to bleed through and the dust/spores to float in the atmosphere, the feeling of emptiness conveyed through those elements in mind. Other features that contributed towards the rendering was LUT (Look up Table) for colour grading and Post Processing effects. I used a LUT system that was free called ColorRama. It’s a fantastic plugin for UE4 that allowed me to change the colour scheme of the environment, allowing me to develop a cinematic effect. The Post-processing effects were done during the end of the environments development. As with many people, Post-processing is just too much fun! I added different effects such as a bit of fringe, changing the hues, increasing Ambient Occlusion intensity, Anti-Aliasing and much more! For some of my renders, Post-processing allowed me to use Field of view, to do close up beauty shots which are nice in some situations as it adds a depth of realism. The lighting overall that I have designed ensure it picks on the dust and speculars on the floor, it did take a long time to build, I also put that down to the amount of foliage meshes I have as well, which are set to the lowest possible lighting resolution to retain performance. I wanted the lighting to highlight dirty surfaces, dented woods, dust and all that good stuff. Good lighting makes a huge impact.
Lessons
I think that having a lot of patience was something I learnt from this. The project had a really rocky start. There were times I was thinking this is not going to work and then I’d have to take a step back and rethink what I am doing. So really take the time to plan it out because it will come together. Avoid playing around with processing effects too quickly as well. This can really hinder you. It’s a lot of fun to play around with post processing effects but it will make you hit a huge road block later on. Allow the creative juices to flow through what you’re trying to achieve first and then enhance it all with rendering effects. It’s a big environment; I got lost quite a few times trying to decide where to go with it. Originally, this environment was going to be just one single screenshot render but I couldn’t have that, I had to make it a whole environment. Something I also learned, no matter how big your project is, keep going and keep striving to push yourself beyond what you can do. You can really develop yourself more if you are willing to push yourself and do something that is hard. I was scared at times to model things I thought I just couldn’t model, in the end I had to believe in myself that I could pull this project off. Every artist out there no matter what skill level you are, you are capable of doing what you are dreaming about doing.
The original reference only had one view and that meant I had to do a lot of improvisation to dress the other two rooms as you can’t see them. I spent quite a few hours of research into the models I would use in those rooms and how they would look. That was a good challenge to make those rooms up of the top of my head and to get them the way I want. I think I painted way too much foliage and this really hindered baking times, it took quite a while to bake. I was also battling UE4 everyday with build errors, was due to an older version I was using, I had to upgrade and then convert my project over to the latest UE4 to get things working again. I really learnt and enjoyed foliage and lighting. I think I have improved my lighting skills. I got an eye for atmosphere over time in this project and I really liked experimenting with how things looked and felt.
The floor was another challenge; I wanted a floor that had dirt, moss, bits of dilapidated wood and grit. I think I pulled it off. It was quite tedious. I actually had to redo it twice as one morning I went to work on my project, to my horror, UE4 project had corrupted and this was a setback. So advice to newcomers and anyone alike, back your work up!! With this project, I think I have learnt about acoustics of an environment, atmosphere and how to convey a story. I tried to convey a story through the assets and lighting and that’s something I’d like to try and keep on developing when I make my next projects. One feature I wanted badly was puddles and like I said I am not a shader expert in UE4. I tried to learn from different sources but unfortunately I couldn’t grasp how to make a wet map, this is something I need to learn eventually, it would have looked great. Glass was also a pain to render properly in UE4. A really awesome glass shader from M_Orzelek on the UE4 forums shared his custom dirt/cracked glass shader which solved most problems with glass but it had some refraction issues.
If I could give any advice to anyone it would be to have fun! Have fun while you are building your environment. I think if you are not having fun making something or not enjoying it, it won’t turn out as good as you want it to because of high expectations. Start by blocking the bare necessities out. If it’s a building or like mine its inside, start blocking out the big shapes, then when you’re happy start blocking out the smaller, more intricate pieces. Keep everything consistent and a tight control on quality. Also, what I emphasized earlier is to have plenty of courage. Have the courage to build and craft what you want to make and it will happen. You will make an amazing piece of work. Try and keep most things modular as possible to ease the production. Also maps, keep maps at a reasonable resolution for certain things such as big detailed things, have them at 2k maps, smaller less detailed props, 256 – 512 maps and use polygons where you most need them. There is more objects in this environment that I can remember but bear in mind, unwrapping these many models can get tedious. So try and change up the jobs you are doing all the time, it breaks up the repetitiveness. In my final words, keep experimenting and seeing what works for you and keep trying to push the art as much as you possibly can.
Thank you Lvl80 for this chance to show my work to everyone and answer your questions! I hope this has been a good insight into my project. Thanks once again for your time and reading my breakdown of The Last of Us house project. Any questions I am more than happy to answer them.